Multi-level Translocation Events Analysis in Solid-State Nanopore Current Traces.
Xinlong Liu,Zepeng Sun,Wei Liu,Feng Qiao,Li Cui,Jing Yang,Jingjie Sha,Jian Li,Li-Qun Xu
DOI: https://doi.org/10.1109/bibm55620.2022.9995453
2022-01-01
Abstract:Solid-state nanopores have shown impressive performances in several sequencing research scenarios, such as biomolecule conformation detection, biomarker identification, and protein fingerprinting. In all these scenarios, accurate event detection is the fundamental step toward data analysis. Most existing event detection methods use either user-defined thresholds or adaptive thresholds determined automatically by the data. The former class depends heavily on human expertise, which is labor-intensive; the latter appears to be more advanced, however, the setting of threshold parameters is somewhat tricky. Hence, the results are usually inconsistent among different methods. In this paper, we develop a novel event detection method, where the selection threshold is computed following the principle governed by an analytical expression. Unlike other methods, each event’s starting and ending points are located based on the slope rather than picking the first point whose current value goes across the baseline. Moreover, we add a method to determine whether multiple levels are present within each event. We then evaluate the method on two groups of current traces generated by short ssDNA and 48.5kb λ-DNA samples, respectively. The results show that our method performs well on detecting challenging translocation events with relatively low amplitudes, and is also able to accurately locate the starting/end points of each level of the events.